feat: add option to ignore dir content via .ts_noindex file

tagstudio/src/core/constants.py

@@ -13,3 +13,5 @@
 
 TAG_FAVORITE = 1
 TAG_ARCHIVED = 0
+
+TS_FOLDER_NOINDEX: str = ".ts_noindex"

tagstudio/src/core/library/alchemy/library.py

@@ -15,7 +15,6 @@
  and_,
  create_engine,
  delete,
- exists,
  func,
  or_,
  select,
@@ -34,6 +33,7 @@
  TAG_ARCHIVED,
  TAG_FAVORITE,
  TS_FOLDER_NAME,
+ TS_FOLDER_NOINDEX,
 )
 from ...enums import LibraryPrefs
 from .db import make_tables
@@ -362,6 +362,8 @@ def verify_ts_folders(self, library_dir: Path) -> None:
  if not full_ts_path.exists():
  logger.info("creating library directory", dir=full_ts_path)
  full_ts_path.mkdir(parents=True, exist_ok=True)
+ # create noindex file to ignore the folder
+ (full_ts_path / TS_FOLDER_NOINDEX).touch()
 
  def add_entries(self, items: list[Entry]) -> list[int]:
  """Add multiple Entry records to the Library."""
@@ -389,10 +391,10 @@ def remove_entries(self, entry_ids: list[int]) -> None:
  session.query(Entry).where(Entry.id.in_(entry_ids)).delete()
  session.commit()
 
- def has_path_entry(self, path: Path) -> bool:
+ def get_path_entry(self, path: Path) -> Entry | None:
  """Check if item with given path is in library already."""
  with Session(self.engine) as session:
- return session.query(exists().where(Entry.path == path)).scalar()
+ return session.scalar(select(Entry).where(Entry.path == path))
 
  def search_library(
  self,

tagstudio/src/core/utils/refresh_dir.py

@@ -4,7 +4,7 @@
 from time import time
 
 import structlog
-from src.core.constants import TS_FOLDER_NAME
+from src.core import constants
 from src.core.library import Entry, Library
 
 logger = structlog.get_logger(__name__)
@@ -39,7 +39,11 @@ def save_new_files(self) -> Iterator[int]:
  self.files_not_in_library = []
 
  def refresh_dir(self, lib_path: Path) -> Iterator[int]:
- """Scan a directory for files, and add those relative filenames to internal variables."""
+ """Scan a directory for changes.
+
+ - Keep track of files which are not in library.
+ - Remove files from library which are in ignored dirs.
+ """
  if self.library.library_dir is None:
  raise ValueError("No library directory set.")
 
@@ -49,24 +53,43 @@ def refresh_dir(self, lib_path: Path) -> Iterator[int]:
  self.files_not_in_library = []
  dir_file_count = 0
 
- for path in lib_path.glob("**/*"):
- str_path = str(path)
- if path.is_dir():
+ for root, _, files in lib_path.walk():
+ if "$RECYCLE.BIN" in str(root).upper():
  continue
 
- if "$RECYCLE.BIN" in str_path or TS_FOLDER_NAME in str_path:
+ # - if directory contains file `.ts_noindex` then skip the directory
+ if constants.TS_FOLDER_NOINDEX in files:
+ logger.info("TS Ignore File found, skipping", directory=root)
+ # however check if the ignored files aren't in the library; if so, remove them
+ entries_to_remove = []
+ for file in files:
+ file_path = root / file
+ entry_path = file_path.relative_to(lib_path)
+ if entry := self.library.get_path_entry(entry_path):
+ entries_to_remove.append(entry.id)
+
+ # Yield output every 1/30 of a second
+ if (time() - start_time_loop) > 0.034:
+ # yield but dont increase the count
+ yield dir_file_count
+ start_time_loop = time()
+
+ self.library.remove_entries(entries_to_remove)
  continue
 
- dir_file_count += 1
- relative_path = path.relative_to(lib_path)
- # TODO - load these in batch somehow
- if not self.library.has_path_entry(relative_path):
- self.files_not_in_library.append(relative_path)
+ for file in files:
+ path = root / file
+ dir_file_count += 1
+
+ relative_path = path.relative_to(lib_path)
+ # TODO - load these in batch somehow
+ if not self.library.get_path_entry(relative_path):
+ self.files_not_in_library.append(relative_path)
 
- # Yield output every 1/30 of a second
- if (time() - start_time_loop) > 0.034:
- yield dir_file_count
- start_time_loop = time()
+  # Yield output every 1/30 of a second
+  if (time() - start_time_loop) > 0.034:
+  yield dir_file_count
+  start_time_loop = time()
 
  end_time_total = time()
  logger.info(

tagstudio/tests/macros/test_refresh_dir.py

@@ -3,6 +3,7 @@
 
 import pytest
 from src.core.enums import LibraryPrefs
+from src.core.library import Entry
 from src.core.utils.refresh_dir import RefreshDirTracker
 
 CWD = pathlib.Path(__file__).parent
@@ -22,3 +23,37 @@ def test_refresh_new_files(library, exclude_mode):
 
  # Then
  assert registry.files_not_in_library == [pathlib.Path("FOO.MD")]
+
+
+@pytest.mark.parametrize("library", [TemporaryDirectory()], indirect=True)
+def test_refresh_removes_noindex_content(library):
+ # Given
+ registry = RefreshDirTracker(library=library)
+
+ # create subdirectory with .ts_noindex file in it
+ (library.library_dir / "subdir").mkdir()
+ (library.library_dir / "subdir" / ".ts_noindex").touch()
+ # add entry into library
+ entry = Entry(
+ path=pathlib.Path("subdir/FOO.MD"),
+ folder=library.folder,
+ fields=library.default_fields,
+ )
+ library.add_entries([entry])
+
+ # create its file in noindex directory
+ (library.library_dir / entry.path).touch()
+ # create another file in the same directory
+ (library.library_dir / "subdir" / "test.txt").touch()
+
+ # add non-ignored entry into library
+ (library.library_dir / "root.txt").touch()
+
+ # When
+ list(registry.refresh_dir(library.library_dir))
+
+ # Then
+ # file in noindex folder should be removed
+ assert not library.get_path_entry(entry.path)
+ # file in index folder should be registered
+ assert registry.files_not_in_library == [pathlib.Path("root.txt")]

tagstudio/tests/test_library.py

@@ -18,11 +18,11 @@ def test_library_add_file(library):
  fields=library.default_fields,
  )
 
- assert not library.has_path_entry(entry.path)
+ assert not library.get_path_entry(entry.path)
 
  assert library.add_entries([entry])
 
- assert library.has_path_entry(entry.path)
+ assert library.get_path_entry(entry.path) is not None
 
 
 def test_create_tag(library, generate_tag):